JP2000357999A - Decoding device, method therefor and program providing medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a decoding device, its method and a program providing medium, by which continuing property as a sound signal is kept and sound with high quality is decoded. SOLUTION: A convolution decoder 16 convolves and decodes a convolution encoding output from the side of an encoding device. An important bit group where an error inspecting code is added and kept as is and a bit group which excludes the important bit group are made to be a convolution decoding output. A CRC code comparison and frame mask part 15 compares a CRC inspecting code added to the convolution decoding output from the decoder 16 with a CRC differential inspecting code calculated from the bit group which excludes the important bit one and adjusts the convolution decoding output.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a decoding apparatus and method for decoding an encoding parameter selectively protected against an error occurring on a transmission path, and a program providing medium.

[0002]

2. Description of the Related Art There are known various encoding methods for compressing an audio signal (including a voice signal and an acoustic signal) by utilizing a statistical property in a time domain and a frequency domain and a characteristic of human perception. ing. As this coding method, VSEL which is a coding method of a so-called CELP (Code Excited Linear Prediction) coding system is used.
P (Vector Sum Excited Linear Prediction) encoding method, PSI-CELP (Pi
tch Synchronus Innovation-CELP (Pitch Synchronous Noise Excitation Source-CELP) coding scheme and the like have recently attracted attention as low bit rate speech coding schemes.

In a waveform coding method such as the CELP coding method, a predetermined number of samples of an input voice signal are divided into blocks or frames as a coding unit, and synthesized into a voice time axis waveform for each block or frame. Vector quantization of the waveform is performed by performing a closed loop search for an optimal vector using an analysis by synthesis method, and the index of the vector is output.

Incidentally, the low bit rate, for example, 2
The coded bits obtained by the kbps or 4 kbps audio coding method are widely applied to fields such as communication, computers, and broadcasting as general audio regardless of specific audio information content. There is a need to protect against errors that occur.

[0005] When an error occurs continuously in the transmission path, a lack of sound or the like continues for a long time at the time of speech decoding, which leads to a decrease in speech quality.

Accordingly, the present applicant has filed Japanese Patent Application No. 09-285.
According to the specification of Japanese Patent No. 903, it is strong against errors occurring in the transmission path,
An encoding method and apparatus and a decoding method and apparatus that can improve quality have been clarified.

In this encoding method and apparatus, an input audio signal is divided in a predetermined encoding unit on a time axis, encoding is performed in each encoding unit, and a plurality of types of audio encoding parameters are output. An important bit group that is important for hearing is selected from a plurality of types of speech coding parameters, an error check code is generated from the important bit group, and convolutional coding is performed on the error check code and the important bit group. Therefore, a group of bits that are important for hearing can be protected from transmission path errors.

In addition, in order to decode the encoded data transmitted by being convolved with the error check code and the important bit group and concatenated with the bit group excluding the important bit group, the decoding method The apparatus performs convolutional decoding on the convolutionally coded output, and sets the important bit group to which the error check code is added and the bit group excluding the important bit group as a convolutional decoded output, and performs convolutional decoding. A transmission error is checked using the error check code added to the decoded output, the convolutional decoded output is adjusted according to the error check result, and a speech decoding process is performed on the adjusted convolutional decoded output. Is applied. For this reason, it is possible to decode the speech in which the deterioration in quality due to the transmission path error is suppressed.

[0009]

By the way, in the above-mentioned decoding method and apparatus, the decoded output is output in accordance with an error check result obtained by checking a transmission error using the error check code added to the convolutional decoded output. At the time of adjustment, it is necessary to maintain continuity as an audio signal after being subjected to audio decoding processing.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and has as its object to provide a decoding apparatus and method capable of maintaining high continuity as an audio signal and decoding high quality audio, and a program providing medium. And

[0011]

In order to solve the above-mentioned problems, a decoding apparatus according to the present invention encodes an input speech signal by dividing the input speech signal by a predetermined coding unit on a time axis. In a decoding device that decodes coding parameters that have been transmitted in a class according to the perceptual sensitivity of bits to transmission path errors, an error check code added to the coding parameters of a predetermined class is used. Transmission line decoding means for detecting an error in the coding unit and making a frame mask process different according to the error of the coding parameter generated in the coding unit.

Here, the encoding processing in the encoding apparatus is performed by using a plurality of vector quantizers, and the index representing each code vector is determined based on the sensitivity of the bit to the transmission path error in terms of auditory sensitivity. The transmission path decoding means detects an error check code added to the index of a predetermined class, and transmits the error check code generated in the coding unit. In accordance with the frame mask processing.

The encoding process in the encoding device is performed using a multi-stage vector quantizer created by alternate learning, and the index representing each code vector is determined by the bit audibility of a transmission line error. By being classified and transmitted according to the above sensitivity,
The transmission path decoding means detects an error check code added to the index of a predetermined class, and varies a frame mask process depending on whether an error is detected in a plurality of indexes.

In order to solve the above-mentioned problems, a decoding method according to the present invention is characterized in that an input audio signal is divided and encoded in a predetermined encoding unit on a time axis by an encoding device, and a bit for a transmission path error is encoded. A decoding method for decoding coding parameters that have been transmitted in a class according to the perceived sensitivity of an object, wherein an error is detected using an error check code added to the coding parameter of a predetermined class. And a transmission line decoding step of changing a frame mask process according to an error of the encoding parameter generated in the encoding unit.

The encoding process in the encoding device is performed by using a plurality of vector quantizers, and an index representing each code vector has an effect on the audibility of a bit to a transmission path error. The transmission path decoding step detects an error check code added to the index of a predetermined class and transmits the error check code to the index error generated in the coding unit. The frame mask processing is changed accordingly.

The encoding processing in the encoding apparatus is performed using a multi-stage vector quantizer created by alternate learning, and the index representing each code vector is determined by the bit audibility of a transmission path error. By being classified and transmitted according to the above sensitivity,
The transmission path decoding step detects an error check code added to the index of a predetermined class, and varies a frame mask process depending on whether an error is detected in a plurality of indexes.

In order to solve the above-mentioned problems, a program providing medium according to the present invention is arranged such that an input device separates and encodes an input speech signal by a predetermined coding unit on a time axis, and that a coding error is reduced. In a program providing medium for providing a decoding program for decoding an encoding parameter transmitted by being classified according to the auditory sensitivity of a bit, the encoding parameter is added to a predetermined class of the encoding parameter. Provided is a program including a transmission line decoding step of detecting an error using an error check code and making a frame mask process different according to the error of the encoding parameter generated in the encoding unit.

The encoding process in the encoding device is performed by using a plurality of vector quantizers, and the index representing each code vector is determined by the perceptual sensitivity of bits to transmission path errors. The transmission path decoding step detects an error check code added to the index of a predetermined class, and detects an error of the index generated in the coding unit. A program is provided for making the frame mask processing different depending on the situation.

The encoding process in the encoding device is performed by using a multi-stage vector quantizer created by alternate learning, and the index representing each code vector is determined by the audibility of the bit with respect to the transmission path error. By being classified and transmitted according to the above sensitivity,
The transmission line decoding step provides a program for detecting an error check code added to the index of a predetermined class and for making a frame mask process different depending on whether an error is detected in a plurality of indexes.

[0020]

Embodiments of the present invention will be described below with reference to the drawings. This embodiment is a mobile phone device shown in FIG. 1 including a decoding device as a specific example of the decoding device and method of the present invention. This portable telephone device is also provided with an encoding device that performs audio encoding at a low bit rate of, for example, 2 kbps or 4 kbps.

The coding apparatus generates a plurality of types of voice coding parameters by dividing the input voice signal into predetermined coding units on the time axis and performing coding for each coding unit. Then, the bits are assigned to a plurality of protection classes according to the audibility of the bits to the transmission path error of the speech coding parameter. For example, when the bit rate is 2 kbps, 6 classes are assigned, and when the bit rate is 4 kbps, 7 classes are assigned. Details of the class assignment and the speech coding parameter will be described later.

First, in this portable telephone device, the coding device includes a speech coder 3, a class division & input order determination unit 23, and a transmission path coder 4. The speech coder 3 divides the input speech signal into predetermined coding units on the time axis, performs coding in each coding unit, and outputs a plurality of types of speech coding parameters. The class division & input order determination unit 23 converts the plurality of types of speech encoding parameters from the speech encoder 3 into 6 when the bit rate is 2 kbps in accordance with the audibility of bits to transmission path errors.
If the class is 4 kbps, the class is divided into seven classes, and the input order to the next stage transmission path encoder 4 is determined. The transmission path encoder 4 includes a class classification & input order determination unit 2
3 and CRC (Cy) according to the input order.
clic Redundancy Check: A check code is generated, multiplied by the CRC, and then protected by convolutional coding, only the CRC is multiplied, or no protection is performed.

The class classification and input order determination unit 23 includes a bit controller 23a and a RAM 23b. The bit controller 23a uses the RAM 23b as a work area for the plurality of types of speech encoding parameters from the speech encoder 3 and classifies them according to the perceptual sensitivity of bits to transmission path errors, and determines the input order.

The transmission line encoder 4 includes a CRC code calculator 5
And a convolutional encoder 6. CRC code calculator 5
Generates a CRC (Cyclic Redundancy Check) check code. The convolutional encoder 6 performs convolutional encoding on the bits to which the CRC check code from the CRC code calculation unit 5 has been multiplied as necessary. Further, the transmission path encoder 4 simply outputs the above C from the CRC code calculation unit 5.
The bit group to which the RC check code is multiplied may be output. In addition, depending on the class, it may pass through without doing anything.

In this portable telephone apparatus, the decoding apparatus to which the decoding apparatus and method according to the present invention are applied performs convolutional decoding on the convolutional coded output, and performs the important check with the error check code added. A convolutional decoder 16 which uses the bit group and the bit group excluding the important bit group as a convolutional decoding output, and the convolutional decoder 1
6 added to the convolutional decoded output from
A CRC code comparing & frame masking unit 15 for comparing the C check code with a CRC error check code calculated from the bit group excluding the important bit group and adjusting the convolutional decoded output according to the comparison result; A speech decoder 17 for subjecting the convolutionally decoded output from the code comparison & frame mask unit 15 to speech decoding processing.

In this portable telephone device, at the time of transmission, an audio signal input from the microphone 1 is transmitted to the A / D converter 2
Is converted into a digital signal by the audio encoder 3, and is converted into an encoding parameter of a low bit rate of 2 kbps / 4 kbps by the audio encoder 3. The class division and input order determination unit 23 determines the classification and input order, and the transmission path encoder After being encoded by 4 so that the quality of the transmission path is hardly influenced by the voice quality, it is modulated by the modulator 7, the output bit is subjected to transmission processing by the transmitter 8, and passed through the antenna duplexer 9.
It is transmitted from the antenna 10.

At the time of reception, a radio wave captured by the antenna 10 is received by the receiver 11 through the antenna duplexer 9, demodulated by the demodulator 13, and a transmission path error is corrected by the transmission path decoder 14. The audio decoder 17 is decoded,
The signal is converted back to an analog audio signal by the D / A converter 18 and output from the speaker 19.

The control unit 20 controls each of the above units, and the synthesizer 12 gives the transmission / reception frequency to the transmitter 8 and the receiver 11. The keypad 21 and the LCD display 22 are used for a man-machine interface.

In the portable telephone device having such a configuration, the CRC code calculation section 5 constituting the transmission path encoder 4 is divided into classes by the class division & input order determination section 23, and the input order is determined. A part or all of a line spectrum pair (LSP) parameter forming an outline of a frequency spectrum of the voice signal, a voiced sound indicating whether the voice signal is voiced (Voice: V) or unvoiced sound (Un Voice: UV) (V) / unvoiced sound (U
V) All of the determination parameters, part or all of the pitch (Pith) parameter when the voice signal is voiced, and linear predictive coding (LPC) residual signal when the voice signal is voiced. Part or all of the spectrum codebook index and gain index indicating the spectrum envelope, and part or all of the noise codebook index and gain index of the linear predictive coding (LPC) residual signal when the speech signal is unvoiced. And generates a CRC from them.

These speech coding parameters are obtained by the speech coder 3. The speech encoding method performed by the speech encoder 3 includes a short-term prediction residual calculation step of calculating a short-term prediction residual of an input audio signal, and a sine wave analysis code for performing sine wave analysis encoding of the determined short-term prediction residual. And a waveform encoding step of encoding the input audio signal by waveform encoding. This speech encoder 3 is shown in FIG.
This will be described with reference to FIG.

The basic concept of the speech encoder 3 shown in FIG. 2 is that a sine wave analysis (sinusoidal analysis) is performed by obtaining a short-term prediction residual of an input speech signal, for example, an LPC (linear predictive coding) residual.
lysis) encoding, such as harmonic coding (har
monic coding), and a second encoding unit 120 that encodes the input audio signal by waveform encoding with phase reproducibility. V: Voiced) first encoding unit 1
The second encoding unit 120 is used to encode the unvoiced (UV) portion of the input signal by using the second encoding unit 120.

The first encoding unit 110 has, for example, L
Harmonic coding and multi-band excitation (M
A configuration for performing sine wave analysis encoding such as BE) encoding is used. The second encoding unit 120 employs, for example, a configuration of code excitation linear prediction (CELP) encoding using vector quantization by closed loop search of an optimal vector using an analysis method based on synthesis.

In the example of FIG. 2, the audio signal supplied to the input terminal 101 is sent to the LPC inverse filter 111 and the LPC analysis / quantization unit 113 of the first encoding unit 110. The LPC coefficient or the so-called α parameter obtained from the LPC analysis / quantization unit 113 is sent to the LPC inverse filter 111, and the LPC inverse filter 111 extracts a linear prediction residual (LPC residual) of the input audio signal. . Also, a quantized output of an LSP (line spectrum pair) is extracted from the LPC analysis / quantization unit 113 and sent to the output terminal 102 as described later. The LPC residual from LPC inverse filter 111 is sent to sine wave analysis encoding section 114. In the sine wave analysis encoding unit 114, pitch detection and spectrum envelope amplitude calculation are performed, and a V (voiced sound) / UV (unvoiced sound) determination unit 1 is performed.
15 is used to determine V / UV. The spectrum envelope amplitude data from the sine wave analysis encoding unit 114 is sent to the vector quantization unit 116. The codebook index from the vector quantization unit 116 as the vector quantization output of the spectrum envelope is
The output from the sine wave analysis encoding unit 114 is sent to the output terminal 104 via the switch 118. Also, the V / UV determination unit 115
Is output to the output terminal 105 and is also sent as a control signal for the switches 117 and 118. In the case of the above-mentioned voiced sound (V), the index and the pitch are selected and each output terminal is output. 103 and 104 respectively.

The second encoding unit 120 in FIG. 2 has a CELP (code excitation linear prediction) encoding configuration in this example, and outputs an output from the noise codebook 121 by a weighted synthesis filter 122. The synthesized voice signal is sent to the subtractor 123, and the audio signal supplied to the input terminal 101 is extracted from the audio signal obtained through the auditory weighting filter 125. 124, the distance is calculated, and a vector that minimizes the error is searched in the noise codebook 121. The time axis waveform is obtained by using a closed loop search using an analysis by synthesis method. Vector quantization is performed. This CELP coding is used for coding unvoiced sound portions as described above, and a codebook index as UV data from the noise codebook 121 is a V / UV determination result from the V / UV determination unit 115. Is output from the output terminal 107 via a switch 127 that is turned on when the sound is unvoiced (UV).

FIG. 3 shows the speech encoder 3 shown in FIG.
FIG. 3 is a diagram showing a more specific configuration of FIG. In FIG. 3, parts corresponding to the respective parts in FIG. 2 are given the same reference numerals.

In the speech encoder 3 shown in FIG. 3, the speech signal supplied to the input terminal 101 has been subjected to a filtering process for removing a signal in an unnecessary band by a high-pass filter (HPF) 109. Then, an LPC analysis circuit 132 of the LPC (linear prediction coding) analysis / quantization unit 113,
It is sent to the LPC inverse filter circuit 111.

The LPC analysis circuit 132 of the LPC analysis / quantization unit 113 obtains a linear prediction coefficient, a so-called α parameter, by an autocorrelation method by applying a Hamming window with a length of about 256 samples of the input signal waveform as one block. .
The framing interval, which is the unit of data output, is 160
Make it about a sample. When the sampling frequency fs is, for example, 8 kHz, one frame interval is 20 for 160 samples.
msec.

The α parameter from the LPC analysis circuit 132 is sent to the α → LSP conversion circuit 133 and is converted into a line spectrum pair (LSP) parameter. This converts the α parameter obtained as a direct type filter coefficient into, for example, ten, ie, five pairs of LSP parameters. The conversion is performed using, for example, the Newton-Raphson method. The conversion to the LSP parameter is because it has better interpolation characteristics than the α parameter.

The LSP parameters from the α → LSP conversion circuit 133 are subjected to matrix or vector quantization by the LSP quantizer 134. At this time, vector quantization may be performed after obtaining an inter-frame difference, or matrix quantization may be performed on a plurality of frames at once. Here, 20 msec is defined as one frame, and LSP parameters calculated every 20 msec are combined for two frames, and are subjected to matrix quantization and vector quantization.

The quantized output from the LSP quantizer 134, that is, the LSP quantization index is supplied to the terminal 102.
And the quantized LSP vector is sent to the LSP interpolation circuit 136.

The LSP interpolation circuit 136 performs the above 20 msec.
Alternatively, the LSP vector quantized every 40 msec is interpolated to make the rate eight times higher. That is, 2.5 mse
The LSP vector is updated every c. This is because when the residual waveform is analyzed and synthesized by the harmonic encoding / decoding method, the envelope of the synthesized waveform becomes a very smooth and smooth waveform.
This is because an abnormal sound may be generated if it changes abruptly every msec. That is, if the LPC coefficient is gradually changed every 2.5 msec, the occurrence of such abnormal noise can be prevented.

In order to perform inverse filtering of the input voice using the LSP vector every 2.5 msec in which such interpolation has been performed, the LSP → α conversion circuit 137
The LSP parameter is converted into, for example, an α parameter which is a coefficient of a direct-order filter of about the tenth order. This LSP → α
The output from the conversion circuit 137 is sent to the LPC inverse filter circuit 111, where the LPC inverse filter 111
Inverse filtering is performed using the α parameter updated every 2.5 msec to obtain a smooth output. An output from the LPC inverse filter 111 is output to an orthogonal transform circuit 145 of a sine wave analysis encoding unit 114, specifically, for example, a harmonic encoding circuit, for example, a DFT.
(Discrete Fourier Transform) sent to the circuit.

The α parameter from the LPC analysis circuit 132 of the LPC analysis / quantization unit 113 is sent to an auditory weighting filter calculation circuit 139 to obtain data for auditory weighting. Vector quantizer 116 and the second encoding unit 12
0 and a synthesis filter 122 with a perceptual weight.

The sine wave analysis encoding unit 114 such as a harmonic encoding circuit analyzes the output from the LPC inverse filter 111 by a harmonic encoding method. That is, pitch detection, calculation of the amplitude Am of each harmonic, determination of voiced sound (V) / unvoiced sound (UV) are performed, and the number of the envelopes or amplitudes Am of the harmonics that change with the pitch is dimensionally converted to a constant number. .

In the specific example of the sine wave analysis encoding unit 114 shown in FIG. 3, general harmonic encoding is assumed. In particular, in the case of MBE (Multiband Excitation) encoding, Modeling is performed on the assumption that a voiced portion and an unvoiced portion exist in the frequency domain at the same time (in the same block or frame), that is, for each band. In other harmonic coding, an alternative determination is made as to whether voice in one block or frame is voiced or unvoiced. In the following description, the term “V / UV for each frame” means that when all bands are UV when applied to MBE coding, the UV of the frame is used. Regarding the MBE analysis / synthesis technique, detailed specific examples are disclosed in the specification and drawings of Japanese Patent Application No. 4-91422 previously proposed by the present applicant.

The open-loop pitch search section 141 of the sine wave analysis encoding section 114 shown in FIG.
01 and the zero-cross counter 1
Signals from the HPF (high-pass filter) 109 are supplied to 42 respectively. The LPC residual or the linear prediction residual from the LPC inverse filter 111 is supplied to the orthogonal transform circuit 145 of the sine wave analysis encoding unit 114. In the open loop pitch search section 141,
An LPC residual of the input signal is used to perform a relatively rough pitch search by an open loop, and the extracted coarse pitch data is sent to a high-precision pitch search 146, and a high-precision closed loop as described later is used. A pitch search (fine search of the pitch) is performed. From the open loop pitch search section 141, a normalized autocorrelation maximum value r (p) obtained by normalizing the maximum value of the autocorrelation of the LPC residual with power together with the coarse pitch data is extracted.
V / UV (voiced sound / unvoiced sound) determination unit 115.

The orthogonal transform circuit 145 performs an orthogonal transform process such as DFT (Discrete Fourier Transform) to convert the LPC residual on the time axis into spectrum amplitude data on the frequency axis. An output from the orthogonal transform circuit 145 is output to a high-precision pitch search unit 146 and a spectrum evaluation unit 148 for evaluating a spectrum amplitude or an envelope.
Sent to

High precision (fine) pitch search section 146
Is supplied with relatively rough coarse pitch data extracted by the open loop pitch search unit 141 and data on the frequency axis, for example, DFT performed by the orthogonal transform unit 145. The high-precision pitch search unit 146 oscillates ± several samples at intervals of 0.2 to 0.5 around the coarse pitch data value to drive the value of the fine pitch data with a decimal point (floating) to an optimum value. At this time, as a method of fine search, a so-called analysis by synthesis method is used, and the pitch is selected so that the synthesized power spectrum is closest to the power spectrum of the original sound. The pitch data from the high-precision pitch search unit 146 by such a closed loop is output via the switch 118 to the output terminal 10.
4

The spectrum evaluation section 148 evaluates the magnitude of each harmonic and a spectrum envelope which is a set of the harmonics based on the spectrum amplitude and the pitch as the orthogonal transform output of the LPC residual, and a high precision pitch search section 146, V / It is sent to a UV (voiced sound / unvoiced sound) determination unit 115 and a vector quantizer 116 with auditory weights.

V / UV (voiced sound / unvoiced sound) determination unit 115
Are the output from the orthogonal transformation circuit 145, the optimum pitch from the high-precision pitch search unit 146, and the spectrum evaluation unit 1
48 and the normalized autocorrelation maximum value r (p) from the open loop pitch search unit 141.
And the V / UV determination of the frame based on the zero cross count value from the zero cross counter 142. Further, the boundary position of the V / UV determination result for each band in the case of MBE may be used as one condition for V / UV determination of the frame. The determination output from the V / UV determination unit 115 is taken out via the output terminal 105.

By the way, an output section of the spectrum evaluation section 148 or an input section of the vector quantizer 116 is provided with a data number conversion (a kind of sampling rate conversion) section. The number-of-data converters are used to make the amplitude data | A _m | of the envelope a constant number in consideration of the fact that the number of divided bands on the frequency axis varies according to the pitch and the number of data varies. It is. That is, for example, if the effective band is up to 3400 kHz, this effective band is divided into 8 bands to 63 bands according to the pitch, and the amplitude data | A _m | of each of these bands is obtained. The number m _MX +1 also changes from 8 to 63. Therefore, the data number conversion unit 119 converts the variable number m _MX +1 of amplitude data into a fixed number M, for example, 4
It is converted into four data.

The above-mentioned fixed number M (for example, 44) of amplitude data or envelope data from the number-of-data conversion section provided at the output section of the spectrum estimating section 148 or the input section of the vector quantizer 116 is used as a vector quantization section. The data is grouped into a vector by a predetermined number, for example, 44 pieces of data, and weighted vector quantization is performed. This weight is given by the output from the auditory weighting filter calculation circuit 139. The index idS of the envelope from the vector quantizer 116
Is taken out of the output terminal 103 via the switch 117. Prior to the weighted vector quantization, an inter-frame difference using an appropriate leak coefficient may be calculated for a vector composed of a predetermined number of data.

Next, the second encoding section 120 will be described. The second encoding unit 120 has a so-called CELP (Code Excited Linear Prediction) encoding configuration, and is particularly used for encoding an unvoiced sound portion of an input audio signal. In this unvoiced CELP coding configuration,
A noise output corresponding to an LPC residual of unvoiced sound, which is a representative value output from a noise codebook, that is, a so-called stochastic codebook 121, is passed through a gain circuit 126 to a perceptual weighted synthesis filter 122. to be sending. Weighted synthesis filter 1
At 22, the input noise is subjected to LPC synthesis processing, and the obtained weighted unvoiced sound signal is sent to the subtractor 123. The subtractor 123 has an HPF from the input terminal 101.
A signal obtained by subjecting the audio signal supplied via the (high-pass filter) 109 to auditory weighting by the auditory weighting filter 125 is input, and a difference or error from the signal from the synthesis filter 122 is extracted. It is assumed that the zero input response of the synthesis filter with auditory weight is subtracted from the output of the auditory weight filter 125 in advance. This error is sent to the distance calculation circuit 124 to calculate the distance,
A representative value vector that minimizes the error is searched in the noise codebook 121. Vector quantization of a time-axis waveform is performed by using a closed-loop search using an analysis by synthesis method.

The data for the UV (unvoiced sound) portion from the second encoding unit 120 using this CELP encoding configuration includes the shape index idSl of the codebook from the noise codebook 121 and the data from the gain circuit 126. The codebook gain index idGl is extracted. The shape index idSl which is UV data from the noise codebook 121 is sent to the output terminal 107s via the switch 127s, and the gain index idGl which is UV data for the gain circuit 126 is sent to the output terminal 107g via the switch 127g. Have been.

Here, these switches 127s, 12s
7g and the switches 117 and 118 are connected to the V / UV
On / off control is performed based on the V / UV determination result from the determination unit 115, and the switches 117 and 118 are turned on when the V / UV determination result of the audio signal of the frame to be currently transmitted is voiced (V). 127s, 127g
Is turned on when the audio signal of the frame to be transmitted at present is unvoiced (UV).

Each parameter output from the speech encoder 3 configured as described above, that is, the LSP parameter LS
P, voiced / unvoiced sound determination parameter VUV, pitch parameter PCH, codebook parameter idS and gain index idG of spectrum envelope, noise codebook parameter idSl and gain index idGl are 2k / 4kb.
FIG. 4 and FIG. 5 show the number of allocated bits separately for ps encoding. FIG. 4 shows parameters common to 2 kbps and 4 kbps. FIG. 5 shows parameters specific to only 4 kbps. Both are parameters per frame.

The LSP parameters are LSP0, LSP2, LSP3,
LSP4 and LSP5. LSP0 is a codebook index of the 10th-order LSP parameter, which is used as a basic parameter of the envelope, and 5 bits are allocated to a 20 msec frame. LSP2 is a codebook index of an LSP parameter for fifth-order low frequency band error correction, and is assigned 7 bits. LSP3 is a codebook index of an LSP parameter for fifth-order high frequency band error correction, and is assigned 5 bits.
LSP5 is a codebook index of an LSP parameter for 10th-order full band error correction, and is assigned 8 bits. Of these, LSP2, LSP3 and LSP5 are indexes used to fill in errors in the previous stage,
In particular, LSP2 and LSP3 are used supplementarily when the envelope cannot be completely expressed by LSP0. The LSP 4 is used to determine whether the encoding mode at the time of encoding is a direct mode (straight mode) or a differential mode (differential mode).
This is a bit selection flag. The selection of a mode having a smaller difference between the LSP of the direct mode obtained by quantization and the LSP obtained by the quantized difference with respect to the original LSP parameter obtained by analysis from the original waveform is shown. When LSP4 is 0, it is in direct mode and LSP4
Is 1 in the difference mode.

The VUV parameter is such that voice encoded data in a predetermined frame is voiced / unvoiced (Voiced / Unvoiced).
, And 2 bits are allocated.

The PCH parameter is a pitch parameter, which is pitch data from the high-precision closed-loop pitch search unit 146 as described above, and is assigned 7 bits.

When the codebook parameter idS of the spectral envelope is 2 kbps, the 0th LPC described by idS0
It is divided into a residual spectrum codebook index and a first LPC residual spectrum codebook index denoted by idS1. In each case, 4 bits are allocated. The 0th and 1st LPC residual spectrum codebook indexes are indices corresponding to different codebooks, and an LPC residual spectrum is formed by adding both codebooks selected therefrom. idG is the LPC residual spectrum gain codebook indesk and is assigned 5 bits. If the codebook parameter idS of the spectrum envelope is 4 kbps, the 0th extended LPC residual spectrum codebook index denoted by idS0_4k and idS1
_4k, a second extended LPC residual spectrum codebook index denoted by idS2_4k, and a third extended LPC residual spectrum codebook index denoted by idS3_4k. Divided into 7 bits are allocated to idS0_4k, 10 bits to idS1_4k, 9 bits to idS2_4k,
Six bits are assigned to idS3_4k.

Here, idS0_4k and others are idS0, idS1, id
G is used to correct the error between the quantized LPC residual spectrum obtained by G and the original LPC residual spectrum.
A correction range is assigned from low frequency components to high frequency components toward _4k.

The other parameters are unvoiced (Unvoic
ed), used for idSL00, idSL11, etc.
Those marked with are the noise codebook index, idGL00
A code with GL such as IDGL11 or idGL11 indicates a noise codebook gain codebook index. Six bits are assigned to idSL00, six bits are assigned to idSL01, four bits are assigned to idGL00, and four bits are also assigned to idGL01. Also, id
SL10 has 5 bits, idSL11 has 5 bits, idSL
5 bits are also allocated to 12 and 5 bits are also allocated to idSL13. Also, idGL10 has 3 bits, idGL1
1 also has 3 bits, idGL12 has 3 bits, idGL13
Are also assigned 3 bits.

Each of the parameters shown in FIGS. 4 and 5 is divided into a plurality of classes by the class division & input order determination unit 23 in accordance with the audibility of bits to transmission path errors.

When the bit rate is 2 kbps (the transmission rate is 3.5 kbps), the class division & input order determination unit 23 divides voiced and unvoiced sounds into, for example, six classes as shown in FIGS. In this figure, the suffix “p” corresponds to the previous frame, and the suffix “c” corresponds to the current frame. That is,
Two frames of the previous frame p and the current frame c are targeted. Also, the smaller the class number, the more significant the bit.

First, the case of a voiced sound will be described with reference to FIG. In the 10th-order codebook desk LSP0 of the LSP parameter, all 5 bits of both the previous frame p and the current frame c are Class I. Also, LSP
In the codebook index LSP2 for the fifth-order low frequency band error correction of the parameter, 2 bits out of 7 bits in both the previous frame p and the current frame c are class I, and the remaining 5
The bit is class VI. Also, a codebook index LS for fifth-order high frequency range error correction of LSP parameters
In P3, one of the five bits in both the previous frame p and the current frame c is set to class I, and the remaining four bits are set to class VI. In addition, the direct mode /
The difference mode selection flag LSP4 is protected by class I for both the previous frame p and the current frame c.

Each of the two bits of the voiced / unvoiced flag VUV is protected by class I for both the previous frame p and the current frame c. In the pitch parameter PCH, 6 bits out of 7 bits are set to class I for both the previous frame p and the current frame c, and the remaining 1 bit is set to class VI.

Further, the 5 bits of the LPC residual spectrum gain codebook index idG are both class I and protected for both the previous frame p and the current frame c. Also, the 0th LPC residual spectrum codebook index idS0 of the previous frame p is all class II, but the 0th LPC residual spectrum codebook index idS0 of the current frame c is all class IV. Also, the first LPC residual spectrum codebook index idS of the previous frame p
All four bits of 1 are class III, but the first LPC residual spectrum codebook index i of the current frame c
All four bits of dS1 are set to class V.

Next, the case where the bit rate is 2 kbps (transmission rate 3.5 kbps) and the voice is unvoiced will be described with reference to FIG. The 10th codebook in-desktop LSP0 of the LSP parameter is the previous frame p and the current frame c.
In all cases, all 5 bits are set to class I.

In the codebook index LSP2 for the fifth-order low frequency band error correction of the LSP parameter, 4 bits out of 7 bits in both the previous frame p and the current frame c are of class I, and the remaining 3 bits are in class VI. It is said.
In the codebook index LSP3 for the fifth-order high frequency band error correction of the LSP parameter, 2 bits out of 5 bits in both the previous frame p and the current frame c are set to class I, and the remaining 3 bits are set to class VI. . Also, LSP
Parameter direct mode / differential mode selection flag LSP4
Is protected by class I for both the previous frame p and the current frame c.

Further, each of the two bits of the voiced / unvoiced flag VUV is protected by class I for both the previous frame p and the current frame c.

Also, the 4 bits of the noise codebook gain codebook index idGL00 are both class I and protected for both the previous frame p and the current frame c. Also, the noise codebook gain codebook index id
The 4 bits of GL01 are also set to Class I for both the previous frame p and the current frame c and protected. Also, the 6 bits of the noise codebook index idSL00 are all set to class VI for both the previous frame p and the current frame c. Also, the 6 bits of the noise codebook index idSL01 are also all set to class VI for both the previous frame p and the current frame c.

In this unvoiced sound, some of the bits of class VI are protected as bits of class II to V,
If an error is detected, no action is taken as in the case of other class VI bits.

When the bit rate is 4 kbps (the transmission rate is 6.2
In the case of (kbps), the class division & input order determination unit 23 divides the voiced sound and the unvoiced sound into, for example, seven classes as shown in FIGS. Also in this figure, the suffix “p” is the previous frame,
The suffix "c" corresponds to the current frame. That is, two frames of the previous frame p and the current frame c are targeted. Similarly, the smaller the class number, the more significant the bit.

First, the case of a voiced sound will be described with reference to FIG. In the 10th-order codebook desk LSP0 of the LSP parameter, all 5 bits of both the previous frame p and the current frame c are Class I. Also, LSP
In the codebook index LSP2 for the fifth-order low frequency band error correction of the parameter, 4 bits out of 7 bits in both the previous frame p and the current frame c are set to class I, and the remaining 3
The bits are assigned to class VII. Also, a codebook index for fifth-order high frequency range error correction of LSP parameters
In LSP3, one of the five bits in both the previous frame p and the current frame c is set to class I, and the remaining four bits are set to class VII. Further, the direct mode / difference mode selection flag LSP4 of the LSP parameter is protected by class I for both the previous frame p and the current frame c.

The two bits of the voiced / unvoiced flag VUV are protected by class I for both the previous frame p and the current frame c. In the pitch parameter PCH, 6 bits out of 7 bits are set to class I for both the previous frame p and the current frame c, and the remaining 1 bit is set to class VII.

Also, the 5 bits of the LPC residual spectrum gain codebook index idG are all set to class I for both the previous frame p and the current frame c and protected. Also, all 4 bits of the 0th LPC residual spectrum codebook index idS0 of the previous frame p are set to class III, but all 4 bits of the 0th LPC residual spectrum codebook index idS0 of the current frame c are set to class V. . Also, all 4 bits of the first LPC residual spectrum codebook index idS1 of the previous frame p are of class IV, but all 4 bits of the first LPC residual spectrum codebook index idS1 of the current frame c are of class VI.
It is said.

Further, 5 bits out of 7 bits of the 0th extended LPC residual spectrum codebook index idS0_4k are set to class I for both the previous frame p and the current frame c, but the remaining 2 bits are set to class VII. . Also, the first
Extended LPC residual spectrum codebook index idS1_4
One of the 10 bits of k is set to class I for both the previous frame p and the current frame c, while the remaining 9 bits are set to class II. One of the 9 bits of the second extended LPC residual spectrum codebook index idS2_4k is set to class I for both the previous frame p and the current frame c, but the remaining 8 bits are set to class II. Also,
Third extended LPC residual spectrum codebook index idS
One bit out of the 6 bits of 3-4k is set to class I for both the previous frame p and the current frame c, but the remaining 5 bits are set to class II.

Next, a case where the bit rate is 4 kbps (transmission rate of 6.2 kbps) and the voice is unvoiced will be described with reference to FIG. The 10th codebook in-desktop LSP0 of the LSP parameter is the previous frame p and the current frame c.
In all cases, all 5 bits are set to class I.

In the codebook index LSP2 for the fifth-order low frequency band error correction of the LSP parameter, 4 bits out of 7 bits in both the previous frame p and the current frame c are class I, and the remaining 3 bits are class VII. It is said.
In the codebook index LSP3 for the fifth-order high frequency band error correction of the LSP parameter, one of the five bits in both the previous frame p and the current frame c is set to class I, and the remaining four bits are set to class VII. . Also, LS
P parameter direct mode / differential mode selection flag LSP
4 is protected by class I for both the previous frame p and the current frame c. Also, the codebook index LSP5 of the LSP parameter of the 10th-order whole band error correction is that one of the 8 bits in both the previous frame p and the current frame c has a class
I is set, and the remaining 7 bits are set to class VII.

The two bits of the voiced / unvoiced flag VUV are protected by class I for both the previous frame p and the current frame c.

Further, the 4 bits of the noise codebook gain codebook index idGL00 are both class I and protected for both the previous frame p and the current frame c. Also, the noise codebook gain codebook index id
The 4 bits of GL01 are also set to Class I for both the previous frame p and the current frame c and protected.

The noise codebook index idSL0
The 6 bits of 0 are all class VII for the previous frame p and the current frame c. Also, the 6 bits of the noise codebook index idSL01 are also set to class VII for both the previous frame p and the current frame c.

The three bits of the noise codebook gain codebook index idGL10 are both class I and protected for both the previous frame p and the current frame c. Also, the noise codebook gain codebook index idGL11
Of the previous frame p and the current frame c are all class I and protected. In addition, the 3 bits of the noise codebook index idGL12 are also set to class I for both the previous frame p and the current frame c. Two bits of the three bits of the noise codebook index idGL13 are all class I for both the previous frame p and the current frame c, but the remaining one bit is class V11.

The noise codebook index idSL1
The five bits of 0 are all class VII for both the previous frame p and the current frame c. Also, the 5 bits of the noise codebook index idSL11 are also set to class VII for both the previous frame p and the current frame c. In addition, the 5 bits of the noise codebook index idSL12 are also set to class VII for both the previous frame p and the current frame c. Also, the 5 bits of the noise codebook index idSL13 are also set to class VII for both the previous frame p and the current frame c.

Here, part of the bits of class VII in the unvoiced sound are also protected as bits of class II to VI, but no action is taken even if they are erroneous, and like other class VII bits, Will be treated.

Next, the input order of the parameters of voiced and unvoiced sounds of 2 kbps from class I to class VI to the transmission path encoder (channel coder) 4 will be described with reference to FIGS.
This will be described with reference to FIG. The determination of the input order is also performed by the class division & input order determination unit 23 as described above. Here, the order of the bits is determined according to the sensitivity of the bits to the audibility of a transmission line error. The suffix “p” indicates the parameter of the previous frame, and “c” indicates the parameter of the current frame. Bit 0 indicates LSB.

First, in FIG. 10, voiced sound class I
Are input from the first bit of the voiced / unvoiced sound determination flag VUV of the previous frame p to the 0th bit of the voiced / unvoiced sound determination flag VUV of the previous frame p. 0th bit of the direct mode / differential mode selection flag LSP4 of the LSP parameter of p,
Then, the fourth bit, the third bit, the second bit, the first bit, and the zero bit of the LPC residual spectrum gain codebook index idG of the previous frame p
Continues to the th bit. In addition, the LSP of the previous frame
The parameter continues to the fourth bit, the third bit, the second bit, the first bit, and the zeroth bit of the tenth-order codebook in-desktop LSP0. Furthermore, the pitch parameter PCH of the previous frame p is 6
The fifth bit, the fifth bit, the fourth bit,
The third bit, the second bit, and the first bit continue. Up to this point, the NO.
0 to NO. The input order is up to 18. Further, NO. 1
9 from the codebook index LSP2 of the fifth-order low frequency band error correction of the LSP parameter of the previous frame p.
Bit follows and the NO. 20 is the L of the previous frame p
The fourth bit of the codebook index LSP3 for fifth-order high frequency band error correction of the SP parameter follows.
21 is followed by the fifth of the codebook in-desktop LSP2 of the LSP parameter of the previous frame p. Hereinafter, NO. 2
2 to NO. By 43, the parameter of the current frame c is set to the above-mentioned NO. 0 to NO. Input is performed so that the order up to 21 is repeated.

The input order of voiced class II bits (from No. 44 to No. 47) is the sum of the 3rd to 0th bits of the 0th LPC residual spectrum codebook index idS0 of the previous frame p. Four bits are determined as shown in FIG.

The input order of the class III bits of the voiced sound (from No. 48 to No. 51) is the total from the third bit to the 0th bit of the first LPC residual spectrum codebook index idS1 of the previous frame p. Four bits are determined as shown in FIG.

The input order of the voiced class IV bits (from NO. 52 to NO. 55) is
0th LPC residual spectrum codebook index idS0
A total of 4 bits from the third bit to the 0th bit in FIG.
Is determined as shown in FIG.

The input order of the voiced class V bits (from No. 56 to No. 59) is
LPC residual spectrum codebook index of idS1
A total of 4 bits from the third bit to the 0th bit in FIG.
Is determined as shown in FIG.

The input order of voiced class VI bits (from No. 60 to No. 79) is as follows. First, the codebook index LSP2 of the 5th-order low frequency band error correction of the LSP parameter of the previous frame p
Bit, the third bit, the second bit,
It continues to the first bit and the zeroth bit. Up to this point, the NO. 60 to NO. 64
The input order is up to. NO. 65 to NO. Up to 68, the third bit, the second bit, the first bit, and the zero bit of the codebook index LSP3 for the fifth-order high frequency band error correction of the LSP parameter of the previous frame p
The th bit follows. NO. 69 is the 0th bit of the pitch parameter PCH of the previous frame p. NO.
70 to NO. Up to 74, the fourth bit, the third bit of the codebook index LSP2 for the fifth-order low frequency band error correction of the LSP parameter of the current frame c,
The second bit, the first bit, and the zeroth bit follow. NO. 75 to NO. Up to 78, the third bit, the second bit, the first bit, and the zeroth bit of the codebook index LSP3 for the fifth-order high frequency band error correction of the LSP parameter of the current frame c follow. And the last NO. 79 is the 0th bit of the pitch parameter PCH of the current frame c.

Next, in FIG. 11, unvoiced sound class I
Are input from the first bit of the voiced / unvoiced sound determination flag VUV of the previous frame p to the 0th bit of the voiced / unvoiced sound determination flag VUV of the previous frame p. 0th bit of the direct mode / differential mode selection flag LSP4 of the LSP parameter of p,
Then, the third bit, the second bit, the first bit, and the zeroth bit of the noise codebook gain codebook index idGL00 of the previous frame p continue. Further, the third bit, the second bit, the first bit, and the zeroth bit of the noise codebook gain codebook index idGL01 of the previous frame continue. Furthermore, the fourth bit, the third bit, and the second bit of the 10th-order codebook in-desktop LSP0 of the LSP parameter of the previous frame p are set to 0
Continues to the th bit. Up to this point, the transmission path encoder 4
NO. 0 to NO. The input order is up to 15. Further, NO. 16 to NO. Until 19, L of the previous frame p
The sixth bit, the fifth bit, the fourth bit, and the third bit of the codebook index LSP2 for the fifth-order low frequency band error correction of the SP parameter follow. N
O. 20 and NO21 contain the fourth bit and the third bit of the codebook index LSP3 for the fifth-order high frequency band error correction of the LSP parameter of the previous frame p. Hereinafter, NO. 22 to NO. By 43, the parameter of the current frame c is set to the above-mentioned NO. 0 to NO. 2
Input is performed so that the order up to 1 is repeated.

The unvoiced sound class II corresponds to the previous frame p
Codebook index for 5th order low frequency range error correction
A total of 3 bits from the 2nd bit to the 0th bit of LSP2,
Codebook index LS for fifth-order high frequency range error correction
The input order of the second bit of P3 (from NO.44 to N
O. Up to 47).

In the unvoiced sound class III, the first bit of the fifth-order high frequency band error correction codebook index LSP3 of the previous frame p is NO. 48, the 0th bit is NO. The input order becomes 49. Also, N
O. The fifth bit of the noise codebook index idSL00 of the previous frame p is entered in NO. 51 contains the fourth bit.

The unvoiced sound class IV is composed of a total of three bits from the second bit to the zeroth bit of the codebook index LSP2 for the fifth-order low frequency band error correction of the current frame c, and a fifth-order high frequency band error. The input order is the second bit of the correction codebook index LSP3 (from NO. 52 to NO. 55).

In the unvoiced sound class V, the first bit of the codebook index LSP3 for the fifth-order high frequency band error correction of the current frame c is NO. 56, the 0th bit is NO. The input order is 57. Also, N
O. The fifth bit of the noise codebook index idSL00 of the current frame c is entered in NO. 59 contains the fourth bit.

The input order of class VI bits of unvoiced sound (from No. 60 to No. 79) is as follows. First, the noise codebook index of the previous frame p
It continues to the third bit, the second bit, the first bit, and the zeroth bit of idSL00. So far,
When the NO. 60 to NO. The input order is up to 63. NO. 64 to NO. Up to 69, 5 of the noise codebook index idSL01 of the previous frame p
The fourth bit, the fourth bit, the third bit,
The second bit, the first bit, and the zeroth bit follow. NO. 70 to NO. Up to 73, the third bit, the second bit, the first bit, and the zero bit of the noise codebook index idSL00 of the current frame c
The th bit follows. And NO. 74 to NO. 7
9, the fifth bit, the fourth bit, the third bit, the second bit, and the first bit of the current frame c.
The 0th bit and the 0th bit follow.

Next, the input order of each parameter of voiced and unvoiced voices of 4 kbps from class I to class VII to the transmission path encoder (channel coder) 4 will be described with reference to FIGS. 12 to 14 and FIGS. 17 is shown. The determination of the input order here is also performed by the class division & input order determination unit 23 as described above. Here, the order of the bits is determined according to the sensitivity of the bits to the audibility of a transmission line error. Subscript "p"
Indicates the parameter of the previous frame, and “c” indicates the parameter of the current frame. Bit 0 is LSB
Is shown.

FIG. 12 shows the input order of voiced class I. NO. 0 to NO. Up to 65, the input order of a total of 66 bits is determined. First, from the first bit of the voiced / unvoiced sound determination flag VUV of the previous frame p, the value of the voiced / unvoiced sound determination flag VUV of the previous frame p is set to 0.
Following the 0th bit, the 0th bit of the direct mode / differential mode selection flag LSP4 of the LSP parameter of the previous frame p, and the 4th bit of the LPC residual spectrum gain codebook index idG0 of the previous frame p Bit, the third bit, the second bit, the first bit, and the zeroth bit. Further, the fourth bit, the third bit of the 10th-order codebook in-desktop LSP0 of the LSP parameter of the previous frame,
The second bit, the first bit, and the zeroth bit continue. Also, the pitch parameter of the previous frame p
The sixth bit, the fifth bit, the fourth bit, the third bit, the second bit, and the first bit of the PCH continue. Up to this point, the NO. 0 to NO. The input order is up to 18. Further, NO. From 19, the codebook index LS for the fifth-order low frequency band error correction of the LSP parameter of the previous frame p
The sixth bit and the fifth bit of P2 (NO.2)
0), the fourth bit (NO. 21), and the third bit (NO. 22). NO. 23 to NO.
Up to 27, the sixth bit of the 0th extended LPC residual spectrum codebook index idS0_4k of the previous frame p,
The fifth bit, the fourth bit, the third bit, and the second bit follow. In addition, NO. The fourth codebook index LSP3 for the fifth-order high frequency band error correction of the previous frame p is entered in NO. Reference numeral 29 contains the seventh bit of the codebook index LSP5 of the LSP parameter for the 10th-order total band error correction of the previous frame p. In addition, NO. 30 to NO. Up to 32, first, the first extended LPC residual spectrum codebook index idS1
_4k, the ninth bit of the second extended LPC residual spectrum codebook index idS2_4k, the eighth bit of the second extended LPC residual spectrum codebook index idS3_4k
Followed by the fifth bit. Hereinafter, NO. 33 to NO.
By 65, the parameter of the current frame c is set to the above-mentioned NO. 0 to NO. Like repeating the order up to 32
Is entered.

FIG. 13 shows voiced sounds from class II to class V.
Indicates the input order up to. Class II total 44 bits
NO. 66 to NO. The input order is up to 109. First, NO. 66 to NO. Up to 74, the first extended LPC residual spectrum codebook index idS1 of the previous frame
_4k 8th bit, 7th bit, 6th bit, 5th bit, 4th bit, 3rd bit, 2nd bit, 1st bit, The 0th bit follows. NO. 75 to NO. Up to 82, the seventh bit, the sixth bit, the fifth bit, the fourth bit, the third bit, and the second bit of the second extended LPC residual spectrum codebook index idS2_4k of the previous frame The 0th bit, the 1st bit, and the 0th bit follow. NO. 83 to NO. Up to 87, the fourth bit, the third bit, the second bit, the first bit, and the zeroth bit of the third extended LPC residual spectrum codebook index idS3_4k of the previous frame follow. And NO. No. 88 to NO. 10
No. 9 until the above NO. 66 to NO. 22 up to 87
The bits for the current frame c are repeated.

A total of 4 bits of the class III correspond to the NO. 110
From NO. By 113, the 3rd bit, the 2nd bit, the 1st bit, and the 0th bit of the 0th LPC residual spectrum codebook index idS0 of the previous frame p are in order.

The total 4 bits of the class IV are NO. 114 to NO. By 117, the third bit, the second bit, the first bit, and the zeroth bit of the first LPC residual spectrum codebook index idS1 of the previous frame p are arranged in this order.

The total 4 bits of the class V are NO. 118 to NO. By 121, the 3rd bit, the 2nd bit, the 1st bit, and the 0th bit of the 0th LPC residual spectrum codebook index idS0 of the current frame c are in order.

FIG. 14 shows classes VI and VII of voiced sounds.
Indicates the input order. The class VI total 4 bits are NO. 12
2 to NO. By 125, the first LPC of the current frame c
3rd bit, 2nd bit, 1st bit, 0th bit of residual spectrum codebook index idS1
It is the order of the bit.

A total of 34 bits of the class VII correspond to the NO. 12
6 to NO. By 159, it is allocated as follows.
NO. 126 to NO. Up to 128, the codebook index LS of the fifth-order low frequency band error correction of the previous frame p
The second bit, the first bit, and the zeroth bit of P2 are assigned. In addition, NO. 129 to N
O. Up to 132, the third bit, the second bit, the first bit, and the zeroth bit of the codebook index LSP3 for the fifth-order high frequency band error correction of the previous frame p are allocated. In addition, NO. 133 to N
O. By 139, the codebook index LSP5 of the LSP parameter for the 10th-order total band error correction of the previous frame is 6
The fifth bit, the fifth bit, the fourth bit,
The third bit, the second bit, the first bit, and the zeroth bit are assigned. In addition, NO.
No. 140 contains the 0th bit of the pitch parameter PCH of the previous frame. 141 and NO142 have the 0th extended LP
1 of C residual spectrum codebook index idS0_4k
The 0th bit and the 0th bit are assigned. And NO. 143 to NO. By 159, the N
O. 126 to NO. A total of 17 bits up to 142 for the current frame c are repeated.

FIG. 15 shows the input order of unvoiced sound class I. NO. 0 to NO. Up to 65, the input order of a total of 66 bits is determined. First, from the first bit of the voiced / unvoiced sound determination flag VUV of the previous frame p, the value of the voiced / unvoiced sound determination flag VUV of the previous frame p is set to 0.
Following the nth bit, the 0th bit of the direct mode / differential mode selection flag LSP4 of the LSP parameter of the previous frame p, the third bit of the noise codebook gain codebook index idGL00 of the previous frame p, and the like It continues to the second bit, the first bit, and the zeroth bit. Further, the third bit, the second bit, the first bit, and the zeroth bit of the noise codebook gain codebook index idGL01 of the previous frame continue. Furthermore, L of the previous frame p
10th codebook in-desk LSP0 of SP parameter
From the fourth bit, the third bit, the second bit to the zeroth bit. Up to this point, the NO. 0 to NO. The input order is up to 15. Further, NO. 16 to NO. Up to 19, the sixth bit, the fifth bit, the fourth bit, and the third bit of the codebook index LSP2 for the fifth-order low frequency band error correction of the LSP parameter of the previous frame p continue. NO. 20, NO21 contains L of the previous frame p.
Fourth bit of codebook index LSP3 for fifth-order high frequency band error correction of SP parameter, previous frame p
The 7th bit of the codebook index LSP5 of the LSP parameter for the 10th-order full-band error correction follows. Also N
O. 22 to NO. Up to 24, the noise codebook gain codebook index idGL10 of the previous frame p is 2
The 0th bit, the 1st bit, and the 0th bit follow. NO. 25 to NO. Up to 27, the noise codebook gain codebook index of the previous frame p
The second bit, the first bit, and the zeroth bit of idGL11 follow. NO. 28 to NO. Up to 30, the second bit, the first bit, and the zeroth bit of the noise codebook gain codebook index idGL12 of the previous frame p continue.

Further, NO. 31, NO. 32 is followed by the second bit and the first bit of the noise codebook gain codebook index idGL13 of the previous frame p. And, hereinafter, NO. 33 to NO. By 65, the parameter of the current frame c is set to the above-mentioned NO. 0 to NO. Input is performed so as to repeat the order up to 32.

FIG. 16 shows class II and class III of unvoiced sound.
Indicates the input order. Although not described in FIG. 9, a part of class VII is protected by CRC as bits from class II to class VI. Will be treated. First, N
O. Reference numeral 66 denotes the 0th bit of the noise codebook gain code index idGL13 of the previous frame p. Next, NO. 67 to NO. 69 contains 5 of the previous frame p.
Codebook index LSP for next low frequency range error correction
2, the second bit, the first bit, and the zeroth bit follow. NO. 70 to NO. Reference numeral 73 denotes the third bit, the second bit, and the first bit of the codebook index LSP3 for the fifth-order high frequency band error correction of the previous frame p.
The 0th bit and the 0th bit follow. In addition, NO.
74 to NO. Reference numeral 80 denotes the sixth bit, the fifth bit, the fourth bit, the third bit, and the second bit of the codebook index LSP5 of the LSP parameter of the 10th-order total band error correction of the previous frame p. , The first bit, and the zeroth bit follow. Also, N
O. 81 to NO. 86 is followed by the fifth bit, the fourth bit, the third bit, the second bit, the first bit, and the zeroth bit of the noise codebook index idSL00 of the previous frame p. . NO. 87
Has the noise codebook index idSL0 of the previous frame p
The fifth of 1 enters. And, hereinafter, NO. 88 to N
O. By 109, the parameter of the current frame c is set to the above-mentioned NO. 66 to NO. Input is performed so as to repeat the order up to 87.

A total of 4 bits of the unvoiced sound class III is N
O. 110 to NO. 113, the fourth bit of the noise codebook index idSL01 of the previous frame p,
The second bit, the second bit, and the first bit follow.

FIG. 17 shows an unvoiced sound from class IV to class VI.
The order of inputting a total of 46 bits up to I is shown. As described above, although not shown in FIG. 9 above, a part of class VII is protected by CRC as bits from class II to class VI. VI
Treated in the same way as the I bit.

First, the class IV NO. 114 is the 0th bit of the noise codebook index idSL01 of the previous frame p. NO. 115 to NO. 117 includes a noise codebook index idSL1 of the previous frame p.
The fourth bit of 0, the third bit, and the second bit follow.

A total of 4 bits of the class V is set to NO. 118 to NO. 121 is followed by the fourth bit, the third bit, the second bit, and the first bit of the noise codebook index idSL01 of the current frame c.

NO. Of Class VI In 122, the 0th bit of the noise codebook index idSL01 of the previous frame p is entered. NO. 123 to NO. Reference numeral 125 denotes the current frame pc noise codebook index idSL10 4
The third bit, the third bit, and the second bit follow.

A total of 34 bits of the class VII correspond to the NO. 12
6 to NO. By 159, it is allocated as follows.
NO. 126, NO. 127 is the first bit of the noise codebook index idSL10 of the previous frame p;
Contains the th bit. In addition, NO. 128 to NO. 1
32 is the noise codebook index id of the previous frame p
The fourth bit, the third bit, the second bit, the first bit, and the zeroth bit of SL11 follow. In addition, NO. 133 to NO. 137 is followed by the fourth bit, the third bit, the second bit, the first bit, and the zeroth bit of the noise codebook index idSL12 of the previous frame p. NO. 138 to NO. 142 is followed by the fourth bit, the third bit, the second bit, the first bit, and the zeroth bit of the noise codebook index idSL13 of the previous frame p. And NO. 143 to NO. 159
By the above NO. 126 to NO. A total of 17 bits up to 142 for the current frame c are repeated.

The coding parameters for the voiced and unvoiced sounds at a bit rate of 2 kbps are in the input order shown in FIGS. 10 and 11, and the coding parameters for the voiced and unvoiced sounds at a bit rate of 4 kbps are as shown in FIG. 17 are input to the transmission path encoder 4 shown in FIG.

Then, the CRC code calculation unit 5 of the transmission path encoder 4 calculates the CRC shown in the following equation (1) for class I if necessary according to the difference between the bit rates of 2 kbps / 4 kbps and the class classification. By the polynomial, CRC code CRC [i]
Get. In addition, for class II and later, an appropriate CRC
Similarly, a CRC code is obtained using a polynomial.

[0118]

(Equation 1)

However,

[0120]

(Equation 2)

[0121]

(Equation 3)

The CRC code calculation unit 5 performs the operations (1) to (3)
The CRC code CRC [i] is obtained by using the expression up to class I, class II, class IV, and class V at 2 kbps.
Although both voiced and unvoiced sounds are targeted, the present invention aims to adjust the output on the decoding device side in accordance with an error detected according to a CRC code applied to the corresponding class of voiced sound except for class I. And

FIG. 18 shows the total number of bits of the 2 kbps voiced sound from class I to class VI. FIG. 18 also shows the total number of bits of a 4-kbps voiced sound, which will be described later, in classes I to VII.

For class 2 of voiced sound of 2 kbps, C
The RC code calculation unit 5 outputs an audio frame (Audio Fram).
e) 44 to 6-bit CRC code CRC [i], (CRC
parity). When class II is targeted, the CRC code calculation unit 5 calculates 1-bit CRC parity from 4 bits of the audio frame. Class III ~
Similarly, when class V is targeted, a 1-bit CRC parity is calculated from 4 bits of an audio frame. When class VI is targeted, CRC parity is not calculated.

Then, for example, using the CRC code CRC [i] obtained by the above equations (1) to (3) and the input order P [i] shown in FIG. 10, the following equation (4) is obtained. A bit string CVin [i] is created as shown in FIG.

[0126]

(Equation 4)

On the other hand, the CRC code calculator 5 calculates a 6-bit CRC parity from 66 bits of the audio frame for the class I of voiced sound of 4 kbps. When class II is targeted, the CRC code calculator 5 calculates a 6-bit CRC parity from 44 bits of the audio frame. When targeting Class III to Class VI,
Each of the audio frames from 4 bits to 1 bit
Calculate CRC parity. When class VII is targeted, CRC parity is not calculated.

In this case, too, the above equations (1) to (3)
The CRC code CRC [i] obtained by the equation and the above-described FIGS.
Using the input order P [i] shown in FIG. 4, a bit string CVin [i] is created as shown in the above equation (4).

Next, as described above, the convolutional coder 6 applies the convolutional code to the bit group to which the CRC check code supplied from the CRC code calculating section 5 is applied as necessary, that is, the bit string CVin [i]. Is applied. In this encoding apparatus, only the bit string CVin [i] corresponding to the class I bits at the bit rates of 2 kbps and 4 kbps is protected by applying a convolutional code.

The convolutional encoder 6 converts the bit string C corresponding to the class I bits at the bit rates of 2 kbps and 4 kbps.
For Vin [i], a convolutional coding method that does not require a Tail bit, for example, the SRC used in H.223 Annex C
PC (Systematic Rate Compatible Punctured Convolu
Convolutional coding is performed by using a conventional code. The convolutional encoding in the convolutional encoder 6 is performed at a double code rate. In FIG. 18 described above, it is indicated as 8/16. Therefore, the total number of bits of 2 kbps class I is (audio frame 44 bits + CRC
parity 6 bits) × 2 = 100 bits. Also, 4
The total number of kbps class I bits is (audio frame 66 bits + CRC parity 6 bits) × 2 = 1
It becomes 44 bits.

From 2 kbps class II to class V, only the CRC code is applied without performing convolutional coding. Therefore, the total number of bits is (audio frame 4 bits + CRC parity 1 bit) .times.
1 = 5 bits. In class VI, since no CRC code is generated, only 20 bits of the audio frame are the total number of bits.

Since the convolutional coding is not performed and only the CRC code is multiplied from class II to class VI of 4 kbps, the total number of bits of class II is (audio frame 44 bits + CRC parity) as shown in FIG.
6 bits) x 1 = 50 bits, and from class III to class VI, (audio frame 4 bits + CRC)
parity 1 bit) × 1 = 5 bits. Also the class
In VII, since no CRC code is generated, only the 34 bits of the audio frame become the total number of bits.

Therefore, a 2 kbps source coder (so
urce coder) has a bit rate of 3.5 kbps because the total of all classes is 140 bits.

A 4 kbps source coder (source c)
oder) has a bit rate of 6.2 kbps since the total of all the classes is 248 bits.

The class division and input order determination unit 23 classifies the bits in accordance with the audibility of the bits with respect to the error in the transmission path, and is protected by the transmission path encoder 4 according to the class. A plurality of types of encoding parameters are modulated by the modulator 7, transmission processing is performed on output bits by the transmitter 8, and transmitted from the antenna 10 through the antenna duplexer 9.

Next, on the decoding device side, for example, a bit string which is a coded output demodulated from the coding device of another portable telephone device via the antenna 10, the antenna duplexer 9, the receiver 11 and the demodulator 13 Is convolutionally decoded by a convolutional decoder 16, and a CRC code comparison & frame mask unit 15 removes the CRC check code added to the convolutional decoded output from the convolutional decoder 16 and bits excluding the important bit group. A CRC error check code calculated from the group is compared, and the convolutional decoded output is adjusted according to the comparison result.

CRC code comparison & frame mask section 15
Is an error checking step of checking a transmission error using the error check code added to the convolutional decoded output from the convolutional decoding step performed by the convolutional decoder 16, and an error in the error checking step. And an output adjusting step of adjusting the convolutional decoded output according to the inspection result.

In particular, when the CRC codes do not match, that is, when a CRC error is detected, if speech decoding is performed using the data of the frame, the speech quality will be greatly degraded. Depending on the degree
Error protection processing (frame masking) is performed.

The frame masking state of the current frame is updated based on the result of class I CRC decoding.
FIG. 19 shows a state transition diagram by the frame masking process. Each state (from state 0 to state 7) transitions in the direction indicated by the arrow. The transition starts from state 0, where "1" on the transition line indicates the transition direction for an error frame, and "0" indicates the transition direction for an error-free frame.

The following parameter replacement is executed according to the value of the state. In the state without error, the state value becomes 0 and the received voice frame bit is processed. Also,
State 7 indicates the time of return.

For example, when decoding the above LSP parameter, if the state variable state is "state 1" to "state 6," the LSP parameter is replaced with that of the previous frame.

In state 7, that is, at the time of return, if LSP4
If = 0 (the LSP mode is direct type), the LSP parameters are calculated from all LSP codes. If LSP4 = 1
If the LSP mode is a difference type, the LSP parameter is calculated from the following equation (5). In other words, in the difference mode, LS
The LSP parameters from the P0 code are interpolated with those of the previous frame.

[0143]

(Equation 5)

In the above equation (5), LSP _base (n) is L of the base layer.
The SP parameter, LSP _prev (n) is the LSP parameter of the previous frame, and LSP _0th (n) is the value decoded from the LSP 0 code. P is an interpolation coefficient, and as shown in FIG. 20, by the number (frame) of CRC error frames immediately before class I,
It changes every 0.1 from 0.7 to 0.0. For example, if the number of erroneous frames in the past is 0, use P = 0.7 and
LSP parameter LSP _base (n) is 0.7 LSP _prev (n) + 0.3 L
SP _0th (n). LSP2, LSP3 and LSP5 codes are ignored and LSP _base (n) is used as the current LSP parameter.

For example, a mute variable mute for controlling the volume of the output sound is set as shown in FIG. 21 according to the value of the state variable state. For example, if the state variable state is 0, the mute variable mute is set to 1.000. If the state variable state is 6, the mute variable mute is set to 0.000. As the mute variable mute at the state variable state = 7, the average of 1.0 and the mute value of the previous frame is used, but when the average exceeds 0.8, it is replaced with 0.8.

The replacement of the voiced sound parameter and the control of the gain are as follows.

When the voiced / unvoiced sound determination parameter VUV is V, the spectral envelope codebook parameters idS0, idS1,
LPC residual spectrum gain codebook indesks id
G, spectrum parameters for 4 kbps idS0_4k to idS
3-4k is replaced with that of the previous frame. Further, to control the volume of the output sound, the gain level of the harmonic level parameter Am [00..127] of the LSP residual signal is controlled as in the following equation (6). In this equation (6), Am _(org) [i]
Is calculated from the spectral parameters.

[0148]

(Equation 6)

In the state 7, if the immediately preceding frame is unvoiced when the voiced sound is restored, the above (6)
Equation (7) is applied instead of the equation. This is to keep the continuity by suppressing the gain. In other words, this is to suppress waveform disturbance due to a difference between the front and rear spectra.

[0150]

(Equation 7)

Note that the previous frame p and the current frame c of the 0th LPC residual spectrum codebook index idS0, the first LPC residual spectrum codebook index idS1.
For example, at 2 kbps, as shown in FIG. 6 above, each of the four bits is a class other than the class I and the class VI, and is therefore individually protected by one CRC bit from FIG. In the state 0 or the state 7, that is, when returning from the normal state, when the CRC errors of these classes are detected at the same time, the quantized harmonic dimension level parameter Am [00..44] of the fixed dimension becomes the following (8). As shown in the formula, Am is the original to suppress the level in the low frequency range
_{Converted from qnt (org)} [1..44].

[0152]

(Equation 8)

In this equation (8), s [i] is Am _{qnt (org)}
Is a coefficient for suppressing, and is set as shown in FIG.

Also, at 4 kbps, the first extended LPC residual spectrum codebook index idS1_4k, the second extended LPC
Residual spectrum codebook index idS2_4k and third extended LPC residual spectrum codebook index idS3
Most bits of _4k are multiple C as class II bits
Protected by RC bit. If an error is detected in class II, the spectral envelope of the enhancement layer is ignored.

Also, for example, if the VUV determination parameter is UV
, When the state variables state = 1 to 6, the noise codebook gain parameters idGL00, idGL01, 4kbps
The noise codebook gain parameters for idGL10 to idGL13 are replaced with those for the previous frame. In the same case, the noise codebook gain parameters idGL00 and idGL01 may be replaced with the previous frame idGL01, and the noise codebook gain parameters for 4 kbps idGL10 to idGL13 may be replaced with the previous frame idGL13. Noise codebook parameters idSL00, idSL01, 4k
The bps noise codebook parameters idSL10 to idSL13 are generated by generating uniform random numbers in the range of individual bits.

Further, in order to control the volume of the output sound,
The LPC residual signal res [00..159] is gain-controlled as shown in the following equation (9). Here, res [i] in the equation is obtained from the noise code parameter.

[0157]

(Equation 9)

The convolutional coded output subjected to the frame masking processing by the CRC code comparing section & frame masking section 15 in this way is supplied to the speech decoder 17.

The structure of the speech decoder 17 is shown in FIGS. 23 and 24. CRC code comparison unit & frame mask unit 1
5, the codebook index corresponding to the quantized output of the LSP (line spectrum pair) via the terminal 202 is output via the terminals 203, 204 and 205 to the index, pitch, and pitch as the envelope quantized output. VU
The V parameter is extracted respectively, and the terminal 207
, An index as data for UV (unvoiced sound) is extracted. Further, the CRC error signal obtained by performing the CRC check in the CRC code comparing section & frame masking section 15 is sent to the unvoiced sound synthesizing section 220.

The index as the envelope quantized output from the terminal 203 is sent to the inverse vector quantizer 212 and inverse vector quantized, and the spectrum envelope of the LPC residual is obtained and sent to the voiced sound synthesizer 211. . The voiced sound synthesizer 211 synthesizes an LPC (linear predictive coding) residual of the voiced sound portion by sine wave synthesis.
A pitch and V / UV decision output from 5 is also provided. The LPC residual of the voiced sound from the voiced sound synthesis unit 211 is
The signal is sent to the LPC synthesis filter 214. Also, the terminal 20
7 from the unvoiced sound synthesizer 22
The LPC residual, which is the excitation vector of the unvoiced portion, is extracted by referring to the noise codebook. This LPC residual is also sent to the LPC synthesis filter 214. In the LPC synthesis filter 214, the LPC residual of the voiced portion and the LPC residual of the unvoiced portion are subjected to LPC synthesis independently of each other. Alternatively, LPC synthesis processing may be performed on the sum of the LPC residual of the voiced sound part and the LPC residual of the unvoiced sound part. Here, the index of the LSP from the terminal 202 is L
The parameter is sent to the PC parameter reproducing unit 213, and the α parameter of the LPC is extracted.
4 LPC by LPC synthesis filter 214
The audio signal obtained by the synthesis is taken out from the output terminal 201.

FIG. 24 shows a more specific configuration of the speech decoder 17 shown in FIG. In FIG. 24, parts corresponding to the respective parts in FIG.
The same designations are given.

The input terminal 202 is supplied with the vector quantization output of the LSP via the CRC code comparing section & frame masking section 15, that is, the so-called codebook index.

The LSP index is calculated by the inverse vector quantizer 231 of the LSP of the LPC parameter reproducing unit 213.
Is subjected to inverse vector quantization to LSP (line spectrum pair) data, sent to LSP interpolation circuits 232 and 233 and subjected to LSP interpolation processing, and then subjected to LPC (linear) by LSP → α conversion circuits 234 and 235. The α parameter is transmitted to the LPC synthesis filter 214. Here, the LSP interpolation circuit 232 and the LSP → α conversion circuit 234 are for voiced sound (V).
The SP interpolation circuit 233 and the LSP → α conversion circuit 235 are for unvoiced sound (UV). Also, the LPC synthesis filter 214
Separates the LPC synthesis filter 236 for the voiced portion and the LPC synthesis filter 237 for the unvoiced portion. That is, LPC coefficient interpolation is performed independently for voiced and unvoiced parts, and LSPs having completely different properties are interpolated between the transition from voiced to unvoiced and the transition from unvoiced to voiced. To prevent the adverse effects of doing so.

The input terminal 203 is supplied with code index data obtained by weighted vector quantization of the spectrum envelope (Am) via the CRC code comparing section & frame masking section 15, and the input terminal 204 is provided with the CRC terminal. The data of the pitch parameter PCH is supplied via the code comparison unit & frame mask unit 15, and the input terminal 205 is supplied with the V / UV determination data via the CRC code comparison unit & frame mask unit 15.

The vector-quantized index data of the spectral envelope Am from the input terminal 203 is
The data is sent to the inverse vector quantizer 212, subjected to inverse vector quantization, subjected to an inverse transform corresponding to the above-described data number conversion, becomes spectral envelope data, and becomes a sine wave synthesizing circuit of the voiced sound synthesizer 211. 215.

If the inter-frame difference is obtained prior to the vector quantization of the spectrum at the time of encoding, the decoding of the inter-frame difference is performed after the inverse vector quantization, and the number of data is converted to obtain the spectrum envelope. To get the data.

The sine wave synthesizing circuit 215 has an input terminal 2
04 and the V / U from the input terminal 205
V determination data is supplied. Sine wave synthesis circuit 21
5, the LPC inverse filter 11 shown in FIGS.
LPC residual data corresponding to the output from 1 is extracted and sent to the adder 218. The specific method of the sine wave synthesis is disclosed in, for example, the specification and drawings of Japanese Patent Application No. 4-91422 or the specification and drawings of Japanese Patent Application No. 6-198451, which were previously proposed by the present applicant. Have been.

Also, the envelope data from the inverse vector quantizer 212 and the pitch and V / UV judgment data from the input terminals 204 and 205 are combined with a noise synthesis circuit for adding noise in a voiced (V) portion. 216. The output from the noise synthesis circuit 216 is sent to an adder 218 via a weighted superposition addition circuit 217. This is an excitation (Excitati) which is input to the LPC synthesis filter of voiced sound by sine wave synthesis.
When on (excitation, excitation) is made, there is a case where there is a feeling of nasal congestion with a low pitch sound such as a male voice, and the sound quality changes suddenly between V (voiced sound) and UV (unvoiced sound) and feels unnatural. Considering a certain point, the LPC synthesis filter input of the voiced sound portion, that is, the excitation, was considered in consideration of parameters based on the speech coded data, for example, pitch, spectrum envelope amplitude, maximum amplitude in a frame, residual signal level, and the like. Noise is added to the voiced portion of the LPC residual signal.

The addition output from the adder 218 is sent to the voiced sound synthesis filter 236 of the LPC synthesis filter 214 and subjected to LPC synthesis processing to become time waveform data, and further to the voiced sound post filter 238v.
, And sent to the adder 239.

Next, the input terminals 207s and 207s of FIG.
7g, the shape index and the gain index as UV data are supplied, respectively, and the unvoiced sound synthesis unit 2
20. The shape index from the terminal 207s is stored in the noise codebook 22 of the unvoiced sound synthesizer 220.
1, the gain index from the terminal 207g is sent to the gain circuit 222, respectively. Noise codebook 2
21 is a noise signal component corresponding to the LPC residual of the unvoiced sound. The noise signal component has a predetermined gain amplitude in the gain circuit 222 and the windowing circuit 223.
And a windowing process for smoothing the connection with the voiced sound portion is performed.

The output from the windowing circuit 223 is output from the unvoiced sound synthesizing section 220 as the LPC synthesis filter 21.
4 is sent to the synthesis filter 237 for UV (unvoiced sound). The synthesis filter 237 performs LPC synthesis processing to obtain unvoiced sound time waveform data. The unvoiced sound time waveform data is filtered by the unvoiced sound post filter 238u, and then sent to the adder 239.

In the adder 239, the time waveform signal of the voiced sound portion from the voiced post filter 238 v and the time waveform data of the unvoiced sound portion from the unvoiced sound post filter 238 u are added and extracted from the output terminal 201.

The audio decoded output extracted from the output terminal 201 is converted into an analog audio signal by the D / A converter 18 and is output from the speaker 19 as audio.

FIGS. 25 and 26 show flowcharts summarizing the encoding method and the audio decoding method performed by the portable telephone device.

That is, the voice coding process executed by the coding device side of the portable telephone device means that the input voice signal is coded in the voice coding step of step S1, and a plurality of types of voice coding from step S1 are performed in step S2. An important bit group having a high audible bit sensitivity to a transmission path error is selected from the parameters, a CRC is calculated from this important bit group, and the CRC and the CRC calculated in step S2 are calculated in step S3. This is a process of performing convolutional encoding on the important bit group.

The speech decoding process executed by the decoding device side of the portable telephone device includes the convolutional decoding of the convolutional coded output from another portable telephone device in step S11 and the CRC in step S12. For the class having the protection bit, error detection is performed by CRC code comparison.
At that time, in step S13, the frame masking state variable state is updated according to the result of the class I error detection. Here, if an error is detected in the class I, the process proceeds to step S14, where frame masking by parameter replacement is performed, and then the process proceeds to step S15, where speech is decoded.

On the other hand, if no error is detected in the class I in step S13, the process proceeds to step S16, and if updated and state = 7, the process proceeds to step S17, where the variable mute is set to the variable mute (p) of all frames. It is determined in step S18 whether or not the result of averaging with 1 exceeds 0.8.
If not, the average is set. If it exceeds 0.8, the process proceeds to step S19, where mute = 0.8 is set.

Next, it is determined in step S20 that LSP4 = 1, and if the mode is the difference mode, the process proceeds to step S21, where an interpolation coefficient is set according to the number of errors of the immediately preceding class I, and LSP decoding is performed in step S22. . At this time, the components by LSP2, LSP3, and LSP5 corresponding to the error are not added. If LSP is not 1 in step S20, the process proceeds to step S23, and normal LSP decoding is performed.

If the state variable state is 0 or 7 and the voiced frame (VUV> 0) in step S24 after the above processing, the flow advances to step S25 to convert the LPC residual spectrum Am into idS
0, idS1 and idG are decoded.
If an error is detected simultaneously from the CRC bits individually added to idS0 and idS1, low-frequency suppression is performed in step S27.

Since Am is converted to a fixed dimension, dimension conversion is performed in step S28 to return to the original dimension based on the pitch.

If the rate is 4 kbps in step S29, the flow advances to step S30 to check the CRC bits covering idS2_4k, idS3_4k, and idS4_4k, which are the extended components of Am, but only when no error is detected as a result. In step S31, an extended component is added.

Further, in step S32, state = 7 and prevVUV = 0 which is the VUV of the previous frame, that is, un
If voiced, the process proceeds to step S33, where Am is suppressed.

As described above, the transmission path encoder 4 and the portable telephone device provided with the encoder 3 as an encoding apparatus on the transmission side can output encoded data that is resistant to errors in the transmission path.

Further, in a portable telephone device having a transmission line decoder to which the decoding method and apparatus according to the present invention is applied and a speech decoder as a decoding device on the receiving side, the quality due to the transmission line error is reduced. It is possible to decode the voice with suppressed reduction.

[0185] Although the portable telephone device provided with the decoding device as a specific example of the decoding device and method of the present invention has been described above, the application of the present invention is not limited only to the decoding device of the portable telephone device. . For example, the present invention can be applied to a transmission system.

FIG. 27 shows a transmission system to which the present invention is applied (a system refers to a logical collection of a plurality of devices, irrespective of whether the devices of each configuration are in the same housing or not). 1 shows a configuration example of one embodiment.

In this transmission system, the client terminal 63 includes the decoding device, and the server 6
1 is provided. Client terminal 63 and server 61
Is, for example, the Internet or ISDN (Integrated
Service Digital Network), LAN (Local Area Net)
work), PSTN (Public Switched Telephone Networ
k) and the like.

When a request for an audio signal such as a song is made from the client terminal 63 to the server 1 via the network 62, the server 61 encodes the audio signal corresponding to the requested song. The parameters are protected according to the perceptual sensitivity of the bits to transmission path errors on the network 62 and transmitted to the client terminal 63. The client terminal 63 decodes the encoding parameters protected from the transmission path error from the server 61 in accordance with the above-described decoding method, and outputs the decoded parameters as audio from an output device such as a speaker.

FIG. 28 shows an example of the hardware configuration of the server 61 in FIG.

A ROM (Read Only Memory) 71 stores, for example, an IPL (Initial Program Loading) program. CPU (Central Processing U
nit) 72 is, for example, I stored in the ROM 71.
According to the PL program, an OS (Operating System) program stored (recorded) in the external storage device 76 is executed, and further, under the control of the OS, a predetermined application program stored in the external storage device 76 is executed. By doing so, a protection process is performed on the audio signal encoding process and the encoding parameters obtained by the encoding, and a process of transmitting the encoded data to the client terminal 63 is performed. RAM (RandomAccess Memor
y) 73 stores programs and data necessary for the operation of the CPU 72. The input device 74 includes, for example, a keyboard, a mouse, a microphone, and an external interface, and is operated when inputting necessary data and commands. Further, the input device 74 also functions as an interface for receiving an input of a digital audio signal provided to the client terminal 63 from outside. The output device 75 includes, for example, a display, a speaker, a printer, and the like, and displays and outputs necessary information. The external storage device 76 is, for example, a hard disk or the like, and stores the above-described OS and predetermined application programs. In addition, the external storage device 76 also stores data necessary for the operation of the CPU 72 and the like. The communication device 77 performs control necessary for communication via the network 62.

The predetermined application program stored in the external storage device 76 is as shown in FIG.
This is a program for causing the CPU 72 to execute the functions of the speech encoder 3, the transmission path encoder 4, and the modulator 7.

FIG. 29 shows an example of the hardware configuration of the client terminal 63 shown in FIG.

The client terminal 63 is composed of the ROM 81 to the communication device 87, and has basically the same configuration as the server 61 composed of the ROM 71 to the communication device 77 described above.

However, the external storage device 86 includes, as an application program, a program for executing the decoding method according to the present invention for decoding encoded data from the server 61, and other processing as described later. And a program for performing the operation.
By executing these application programs, decoding, reproduction processing, and the like of encoded data protected against transmission path errors are performed.

That is, the external storage device 86 stores an application program for causing the CPU 82 to execute the functions of the demodulator 13, the transmission path decoder 14, and the audio decoder 17 shown in FIG. Have been.

For this reason, the client terminal 63 uses the decryption method stored in the external storage device 86 according to FIG.
Can be realized as software without requiring the hardware configuration shown in FIG.

The client terminal 63 stores the encoded data transmitted from the server 61 in the external storage device 86, reads out the encoded data at a desired time, executes the decoding method, and executes the decoding method. The audio may be output from the output device 85 at the time. Further, the encoded data may be recorded on an external storage device different from the external storage device 86, for example, a magneto-optical disk or another recording medium.

Further, in the above-described embodiment, a recordable medium such as an optical recording medium, a magneto-optical recording medium, and a magnetic recording medium is used as the external storage device 76 of the server 61, and the recording medium is stored in the storage medium. Encoded data may be recorded.

[0199]

According to the decoding apparatus and method of the present invention, high quality speech can be decoded while maintaining continuity as a speech signal.

Further, by using the program providing medium according to the present invention, it is possible to maintain continuity as an audio signal and decode high-quality audio in a computer system.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a mobile phone device according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a basic configuration of a speech encoder constituting the mobile phone device.

FIG. 3 is a block diagram showing a detailed configuration of the speech encoder.

FIG. 4 is a diagram showing parameters common to bit rates of 2 kbps and 4 kbps.

FIG. 5 is a diagram showing parameters specific to only 4 kbps.

FIG. 6 is a diagram illustrating class division of voiced sound parameters at a bit rate of 2 kbps.

FIG. 7 is a diagram showing class division of unvoiced sound parameters at a bit rate of 2 kbps.

FIG. 8 is a diagram showing class division of voiced sound parameters at a bit rate of 4 kbps.

FIG. 9 is a diagram illustrating class division of unvoiced sound parameters at a bit rate of 4 kbps.

FIG. 10 is a diagram showing an input order of voiced sound parameters to a channel coder at a bit rate of 2 kbps.

FIG. 11 is a diagram showing an input order of unvoiced sound parameters to a channel coder at a bit rate of 2 kbps.

FIG. 12 is a diagram showing an input order of class I bits of a voiced sound parameter to a channel coder at a bit rate of 4 kbps.

FIG. 13 is a diagram showing the order of input of voiced sound parameters from class II to class V bits to a channel coder at a bit rate of 4 kbps.

FIG. 14 is a diagram showing the order of input of voiced sound parameter class VI and class VII bits to a channel coder at a bit rate of 4 kbps.

FIG. 15 is a diagram illustrating an input order of class I bits of unvoiced sound parameters to a channel coder at a bit rate of 4 kbps.

FIG. 16 is a diagram showing the order of input of unvoiced sound parameters of class II to class V bits to a channel coder at a bit rate of 4 kbps.

FIG. 17 is a diagram showing an input order of class VI and class VII bits of an unvoiced sound parameter to a channel coder at a bit rate of 4 kbps.

FIG. 18 shows 2 kbps voiced sounds from class I to class VI.
It is a figure which shows the bit assignment of the class II-class VII of the voiced sound of 4 kbps.

FIG. 19 is a state transition diagram for explaining the operation of the transmission line decoder constituting the mobile phone device.

FIG. 20 is a diagram showing a value of an interpolation coefficient P used in the above equation (7).

FIG. 21 is a diagram for explaining setting of a mute variable mute for controlling a volume of an output sound according to a value of a state variable state.

FIG. 22 is a diagram showing a value of s [i] used in the above equation (10).

FIG. 23 is a block diagram showing a basic configuration of the speech decoder.

FIG. 24 is a block diagram showing a detailed configuration of the audio decoder.

FIG. 25 is a flowchart illustrating an encoding method performed by the mobile phone device.

FIG. 26 is a flowchart illustrating a decoding method performed by the mobile phone device.

FIG. 27 is a block diagram of a transmission system to which the present invention can be applied.

FIG. 28 is a block diagram of a server constituting the transmission system.

FIG. 29 is a block diagram of a client terminal constituting the transmission system.

[Explanation of symbols]

Reference Signs List 3 audio encoder, 4 transmission line encoder, 5 CRC code calculation unit, 6 convolutional encoder, 14 transmission line decoder, 15 CRC code comparison unit & frame mask unit, 16
Convolutional decoder, 23 class division & input order determination unit

Claims (13)

[Claims] An encoding apparatus separates and encodes an input speech signal in a predetermined coding unit on a time axis, and classifies the input speech signal in accordance with the audibility of a bit with respect to a transmission path error and transmits the signal. A decoding device that decodes the obtained coding parameter, wherein an error is detected using an error check code added to the coding parameter of a predetermined class, and an error of the coding parameter generated in the coding unit is detected. A transmission path decoding means for making a frame masking process different according to the decoding method. 2. The encoding process in the encoding device is performed using a plurality of vector quantizers.
Since the index representing each code vector is transmitted after being divided into classes according to the perceptual sensitivity of the bit to the transmission path error, the transmission path decoding means allows the transmission path decoding means to add the error added to the index of the predetermined class. 2. The decoding apparatus according to claim 1, wherein a check code is detected, and a frame mask process is changed in accordance with the index error generated in the coding unit. 3. An index transmitted from the encoding device includes: a class to which an error check code is applied according to an audibility of a bit with respect to a transmission path error, and then a convolutional coding is performed; 3. The decoding device according to claim 2, wherein the class is divided into a class that can only be applied and a class that does nothing. 4. The transmission path decoding means performs frame mask processing when an error is detected in a predetermined number or more of coding units, and otherwise uses an erroneous index as it is. Item 3. The decoding device according to Item 2. 5. The encoding process in the encoding device is performed by using a multi-stage vector quantizer created by alternate learning, and an index representing each code vector is used as a bit perception of a transmission path error. The transmission path decoding means detects an error check code added to the index of a predetermined class by being transmitted in a class according to the above sensitivity, and an error is detected in a plurality of indexes. 2. The decoding device according to claim 1, wherein the frame mask processing is changed depending on whether the frame mask processing has been performed. 6. An encoding process in the encoding device,
This is a coding process for quantizing and transmitting audio spectrum information by a two-stage vector quantizer based on alternate learning, and the transmission path decoding means simultaneously detects inaccurate indexes representing code vectors output from the two vector quantizers. 6. The decoding device according to claim 5, wherein a frame mask process for suppressing a low frequency band is performed in a case where the decoding is performed. 7. The transmission line decoding means according to claim 1, wherein when an error is detected from one or more consecutive frames which are predetermined coding units on said time axis, a first normal frame is received. 2. The decoding apparatus according to claim 1, wherein when the frame is a voiced sound and the last normal frame before the error is unvoiced, a frame masking process for suppressing a signal level is performed. 8. An encoding apparatus encodes an input speech signal by dividing the speech signal into predetermined coding units on a time axis, classifying the input speech signal into classes according to the audibility of bits with respect to a transmission path error, and transmitting the class. A decoding method for decoding the coding parameters, wherein an error is detected using an error check code added to the coding parameters of a predetermined class, and the coding parameters generated in the coding unit are detected. A transmission line decoding step of making a frame mask process different according to the error of (1). 9. The encoding process in the encoding device is performed using a plurality of vector quantizers.
Since the index representing each code vector is transmitted after being classified into classes according to the perceptual sensitivity of the bit to the transmission path error, the transmission path decoding step performs the error added to the index of the predetermined class. 9. The decoding method according to claim 8, wherein a check code is detected, and a frame mask process is changed according to the index error generated in the coding unit. 10. The encoding process in the encoding device is performed by using a multi-stage vector quantizer created by alternate learning, and an index representing each code vector is used as a bit perception of a transmission path error. By being transmitted by being classified according to the above sensitivity, the transmission path decoding step detects an error check code added to the index of a predetermined class, and an error is detected in a plurality of indexes. 9. The decoding method according to claim 8, wherein the frame mask processing is made different depending on whether or not the decoding is performed. 11. An encoding apparatus encodes an input audio signal by dividing the input audio signal into predetermined coding units on a time axis and classifying the input audio signal into classes according to the audibility of bits with respect to a transmission path error. A program providing medium for providing a decoding program for decoding the obtained coding parameter, wherein an error is detected using an error check code added to the coding parameter of a predetermined class, and the coding unit A program providing medium, comprising: providing a program having a transmission path decoding step of changing a frame mask process according to an error of the encoding parameter generated in the program. 12. The encoding process in the encoding device is performed by using a plurality of vector quantizers, and an index representing each code vector has a sensitivity to a perceptual sensitivity of a bit to a transmission path error. The transmission path decoding step detects an error check code added to the index of a predetermined class, and detects an error of the index generated in the coding unit. 12. The program providing medium according to claim 11, wherein a program for making a frame mask process different in accordance with the program is provided. 13. The encoding process in the encoding device is performed by using a multi-stage vector quantizer created by alternate learning, and an index representing each code vector is a bit audibility of a bit with respect to a transmission path error. The transmission path decoding step detects an error check code added to the index of a predetermined class by being transmitted in a class according to the above sensitivity, and an error is detected in a plurality of indexes. 12. The program providing medium according to claim 11, wherein a program for making a frame mask process different depending on whether or not the program is executed is provided.